The questions asked of a witness can alter the memory of the witness in surprising ways.
Following that lead, Loftus won funding in 1974 for a proposal to study witness accounts of accidents, and she soon published the first of several influential studies revealing the limitations of eyewitness testimony1. She showed people film clips of car accidents and asked them to estimate the speed of the cars. The wording of the questions, she found, had a profound effect on the estimates. People who were asked, “How fast were the cars going when they smashed into each other?” gave higher estimates on average than those with whom the verb ‘hit’ was used. And those who were told that the cars had ‘contacted’ each other gave the lowest estimates.
Those asked about cars smashing into one another were more than twice as likely as others to report seeing broken glass when asked about the accident a week later, even though there was none in the video. “I realized that these questions were conveying information,” says Loftus. “I began to think of it as a process of memory contamination, and we eventually called it the misinformation effect.” (Source)